PCB Antenna and Feed Resources

S11 Measurement Guide | Calibration & Pitfalls

A practical guide to S11 measurement covering VNA calibration, connector handling, common error sources, and the relationship between S11, VSWR, and return loss—with an interactive calculator and actionable tips for RF PCB work.

VNA connected to an RF PCB for S11 measurement on a lab bench

What Is S11 and Why Does It Matter?

S11 is the input port voltage reflection coefficient of a network, expressed as a complex ratio of reflected wave to incident wave. A perfect 50 Ω load returns S11 = 0 (−∞ dB); a total reflection gives |S11| = 1 (0 dB). On an RF PCB, measuring S11 accurately tells you whether an antenna, filter, or transmission-line transition is matched at the frequencies you care about—and by how much it deviates from the target impedance.

S11 is the most-used single-port S-parameter because it can be captured with one VNA port and one cable, yet it reveals resonant frequency, bandwidth, impedance locus, and quality factor in a single sweep. Return loss (RL) and VSWR are simply different ways of presenting the same data:

  • Return loss (dB) = −20 log₁₀ |S11|. A matched port shows high return loss (e.g., 20 dB means only 1 % of power is reflected).
  • VSWR = (1 + |S11|) / (1 − |S11|). A 1.0 : 1 VSWR is a perfect match; most RF systems target ≤ 2.0 : 1 (S11 ≤ −9.5 dB).
  • Reflection coefficient (Γ) is the linear magnitude of S11, ranging from 0 to 1.

Converting Between S11, VSWR, and Return Loss

SOLT calibration standards for VNA one-port calibration

Engineers constantly hop between these three representations. The relationships are exact and lossless—any one value determines the other two. Enter your known quantity in the calculator below to get the rest instantly.

Loading Reflection VSWR Calculator…

For example, if your VNA reads S11 = −14 dB at the antenna’s center frequency, the calculator returns |Γ| ≈ 0.20, VSWR ≈ 1.50 : 1, and about 4 % reflected power. That is a solid match for most antenna production acceptance criteria, where −10 dB is the typical pass/fail line and −15 dB is a comfortable margin. If your result lands between −10 dB and −6 dB, the mismatch is significant enough to degrade noise figure in a receive chain or waste power in a transmit chain—time to investigate the matching network or connector interface.

Equipment and Setup for S11 Measurement

Vector Network Analyzer Basics

A VNA measures both magnitude and phase of S11 by comparing the reflected signal against a known reference. For one-port S11 work you need only Port 1, a test cable, and a calibration kit. Key specs to check before you start:

  • Frequency range: your sweep must cover at least 20 % beyond the band of interest so you can see out-of-band behavior and resonance tails.
  • Dynamic range: for return-loss measurements beyond −30 dB, you need a VNA with at least 90 dB of receiver dynamic range at the test frequency. Most modern benchtop VNAs (Keysight PNA-L, R&S ZNB, Copper Mountain) exceed 100 dB below 20 GHz.
  • Number of points: 201 or 401 points across the sweep is a practical starting point. Too few points can miss a narrow resonance; too many slows the sweep and can introduce drift errors if the DUT is thermally sensitive.
  • IF bandwidth: narrower IFBW (e.g., 1 kHz vs. 10 kHz) lowers the noise floor but increases sweep time. For most PCB-level S11 work, 1–3 kHz is a good compromise.

Cables and Connectors

The cable between the VNA port and the DUT is the single largest source of measurement error after calibration quality. Use phase-stable test cables rated for your frequency band—typical choices are Sucoflex, Gore Phaseflex, or MegaPhase cables. Flexible lab cables with SMA connectors are fine to 18 GHz; above that, move to 2.92 mm (K) or 2.4 mm connectors to avoid moding.

Connector torque matters. SMA and 3.5 mm connectors require 5 in-lb (0.56 N·m); 2.4 mm connectors require 8 in-lb. Use a calibrated torque wrench every time—finger-tight connections introduce 1–3 dB of return-loss uncertainty above 10 GHz. If your DUT uses an SMA edge-launch connector soldered to the PCB, inspect the solder joint under magnification: a fillet that is too large or too small shifts the reference plane and distorts S11.

VNA Calibration for Accurate S11

Calibration removes systematic errors—directivity, source match, and frequency response—from the measurement path. Without it, the raw S11 trace is dominated by the cable and adapter, not the DUT. A detailed comparison of calibration algorithms is covered in our VNA calibration methods guide; here we focus on the practical steps and the pitfalls specific to S11.

SOLT (Short-Open-Load-Thru) One-Port

For a single-port S11 measurement, you apply three known standards at the end of the test cable:

  1. Short — reflects all energy with a 180° phase shift. Connect the short standard and let the VNA acquire.
  2. Open — reflects all energy with 0° phase shift (plus a small fringing capacitance modeled in the cal-kit definition file).
  3. Load — a precision 50 Ω termination that ideally reflects nothing. The residual return loss of a good cal-kit load is better than −50 dB at low GHz and around −36 dB at 26.5 GHz.

After acquiring all three, the VNA solves for the three one-port error terms and applies correction in real time. The reference plane is now at the connector face where you attached the standards.

Electronic Calibration (ECal)

An ECal module presents multiple impedance states electronically, removing the need to manually swap standards. It is faster and eliminates one major error source—connector wear from repeated mate/de-mate cycles. ECal is especially valuable above 20 GHz where mechanical standards are harder to handle repeatably.

Where the Reference Plane Sits

This is the single most misunderstood concept in S11 measurement. Calibration sets the reference plane at the point where you connected the standards. Everything between that plane and the actual DUT—an adapter, a board-edge connector, a short trace—becomes part of the measured S11. If you calibrate at the cable tip but your DUT’s matching network starts 15 mm down a 50 Ω microstrip, those 15 mm add electrical delay (phase rotation on the Smith chart) that masks the true impedance of the circuit.

Solutions: calibrate as close to the DUT as possible, use port extension (electrical delay) to mathematically shift the plane, or apply TDR-based time-domain gating to window out the connector transition. Each approach has trade-offs in accuracy and convenience.

Common S11 Measurement Pitfalls

Even experienced engineers trip over these. Fixing them often improves measured return loss by 3–10 dB—the difference between a “failed” and a “passing” DUT.

1. Cable Movement After Calibration

Flexing or repositioning the test cable after calibration changes its phase response. On a phase-stable cable, this is typically < 0.5 dB of return-loss error below 6 GHz, but it can exceed 2 dB at 26 GHz on a standard flexible cable. Rule: after calibration, tape or clamp the cable so it does not move during the DUT sweep.

2. Wrong or Missing Cal-Kit Definition

Every calibration kit ships with a definition file (polynomial coefficients for the open’s fringing capacitance, the short’s inductance, and the load’s return loss vs. frequency). Loading the wrong file—or using the VNA’s generic “ideal” standards—introduces systematic offset. At 20 GHz, the open capacitance of a typical 3.5 mm standard is about 50 fF; ignoring it shifts S11 phase by several degrees, which rotates the Smith chart trace and makes impedance matching look worse (or better) than it really is.

3. Adapter and Connector Mismatch

If you calibrate with an SMA female standard but connect to the DUT through an SMA-to-U.FL adapter, that adapter’s return loss (often only −15 to −20 dB above 3 GHz) directly corrupts the measurement. Every adapter between the cal plane and the DUT must be included in the calibration or characterized and de-embedded. This issue is common in insertion loss measurement as well.

4. Ground Loops and Radiation

At frequencies below 1 GHz, ground-loop currents between the VNA, the DUT, and other bench equipment can create ripple on the S11 trace. Above 10 GHz, open fixtures radiate, and nearby metal objects (monitor arms, scope probes) create reflections that modulate S11. Keep the measurement area clear, use absorber foam around the DUT if necessary, and verify repeatability by running the same sweep twice without touching anything.

5. Thermal Drift

VNA receivers and cables drift with temperature. If the lab temperature swings by more than 2 °C during a measurement session, re-calibrate. For long automated test runs, enable the VNA’s internal temperature compensation if available, or schedule periodic re-calibration.

6. Ignoring the Smith Chart

Looking only at the log-magnitude (dB) plot tells you how much energy is reflected but not why. The Smith chart shows whether the mismatch is capacitive or inductive and by how much—information you need to fix the matching network. Always check both views. When correlating measured S11 with an electromagnetic simulation, the Smith chart is the fastest way to spot a systematic offset, as discussed in our measurement-versus-simulation correlation article.

S11 on Antennas: Special Considerations

Smith chart on a VNA display during antenna S11 measurement

Antenna S11 is uniquely sensitive to the measurement environment because the DUT is, by design, an open radiator. A hand near a patch antenna at 2.4 GHz can shift S11 by 3–5 dB at resonance. Best practices:

  • Mount the antenna on a ground plane or fixture that matches its intended installation. A bare PCB antenna measured on a wooden table will show a different S11 than the same antenna in a metal enclosure.
  • Keep the cable routed away from the radiating aperture and behind the ground plane where possible.
  • For absolute antenna performance (gain, efficiency, pattern), S11 alone is not enough—you also need antenna pattern and gain measurements in a calibrated chamber.
  • If you are tuning a prototype, use a jig that lets you repeatedly place the board in the same position so that S11 changes reflect design changes, not fixture variation.

Probe-Based S11 Measurement

When a DUT has no coaxial connector—an MMIC, a bare-die PA, or a PCB test coupon—you measure S11 through a probe station with GSG (ground-signal-ground) probes. Calibration moves to an impedance standard substrate (ISS) with on-wafer short, open, load, and thru patterns, using a technique such as LRRM or multiline TRL. Probe-based S11 can be accurate to 40 GHz and beyond, but probe placement repeatability (±5 µm) and contact resistance become critical error sources.

Post-Processing and Interpreting S11 Data

Time-Domain Gating

Most VNAs can inverse-FFT the frequency-domain S11 into a time-domain reflectometry view. You can then apply a time gate to isolate the DUT’s response from connector and fixture reflections, effectively moving the reference plane without re-calibrating. This is powerful but requires enough frequency span (bandwidth) to achieve the time-domain resolution you need. A rule of thumb: time resolution ≈ 1 / (frequency span), so a 20 GHz span gives roughly 50 ps resolution, corresponding to about 7.5 mm in free space.

De-Embedding

If you have an S-parameter model of the connector or fixture (from simulation or a separate TRL calibration), you can mathematically remove it from the measured data. This is standard practice in MMIC and high-frequency PCB work. The IEEE microwave measurement standards provide frameworks for de-embedding accuracy.

Checking Against Simulation

A measured S11 that disagrees with your HFSS or CST simulation by more than 2–3 dB in return loss or more than a few percent in resonant frequency shift usually points to one of three causes: incorrect material Dk used in simulation (verify against measured Dk/Df data), fabrication tolerance in trace width or dielectric thickness, or a measurement artifact from the issues listed above. Systematically ruling out measurement error first saves hours of unnecessary simulation re-runs.

S11 Measurement Checklist

Print this and tape it next to your VNA.

  1. Inspect all connectors for damage; clean with isopropyl alcohol and lint-free wipes.
  2. Set frequency range, number of points (≥ 201), and IF bandwidth (1–3 kHz).
  3. Perform full one-port calibration (SOLT or ECal) at the end of the test cable, using the correct cal-kit definition file.
  4. Verify calibration: re-measure the load standard—it should read better than −40 dB across the band.
  5. Secure the cable so it will not move during DUT connection.
  6. Connect the DUT with a torque wrench (5 in-lb for SMA/3.5 mm, 8 in-lb for 2.4 mm).
  7. Capture S11 in both log-magnitude and Smith chart views.
  8. Save the Touchstone (.s1p) file, not just a screenshot—you will need complex data for de-embedding or simulation comparison.
  9. If results look suspicious, disconnect and reconnect the DUT to check repeatability before changing the design.

Frequently Asked Questions

What is a good S11 value for an RF PCB antenna?

Most antenna specifications require S11 ≤ −10 dB (VSWR ≤ 1.92 : 1) across the operating bandwidth. High-performance designs often target −15 dB or better at center frequency to allow margin for production variation and environmental detuning.

Can I measure S11 with a scalar network analyzer or spectrum analyzer?

A scalar analyzer or a spectrum analyzer with a tracking generator can show |S11| magnitude but not phase. Without phase, you cannot plot the Smith chart, perform time-domain gating, or de-embed fixtures. For serious RF PCB work, a VNA is essential. Our VNA insertion loss testing overview covers two-port extensions of the same principles.

How often should I re-calibrate during a measurement session?

Re-calibrate whenever you change the cable, adapter, or frequency range, or if the ambient temperature has shifted by more than 2 °C. For automated production testing, a 30-minute to 1-hour re-calibration interval is common practice.

Does PCB material affect S11 measurement accuracy?

The material does not affect measurement accuracy directly, but it affects the DUT’s actual S11. A laminate with tighter Dk tolerance (e.g., ±0.02 for Rogers RO4350B vs. ±0.15 for generic FR-4) produces more predictable impedance and therefore more repeatable S11 across a production lot.

What is the difference between S11 and S22?

S11 is the reflection coefficient at Port 1; S22 is the reflection coefficient at Port 2. For a symmetric two-port network (e.g., a straight 50 Ω transmission line), S11 ≈ S22. For an asymmetric device (e.g., an amplifier), they differ and both must be measured. S11 and S22 are both one-port reflection measurements; the only difference is which port you excite.

Ready for a quotation?

Send your requirements for a material, stackup and manufacturing review. Price and schedule are confirmed for your project.

Upload Gerber